Arguments against helmet legislation are flawed
BMJ 2006; 332 doi: https://doi.org/10.1136/bmj.332.7543.725 (Published 23 March 2006) Cite this as: BMJ 2006;332:725
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Hagel et al have failed to grasp the main point of Robinson's
analysis. Regardless of how effective a particular safety measure might
be in theory, or even empirically from case-controlled studies, a failure
to demonstrate any real benefits in whole populations must necessarily
call into question its true effectiveness--particularly where compulsion
may be involved.
While helmet proponents are often quick to claim any reduction in
cycling fatalities as being primarily due to increased helmet use, they
have thus far been unable to demonstrate that these reductions are not
better explained by fewer numbers of cyclists--due at least in part to the
imposition of a mandatory helmet law (MHL) and/or other traffic safety
measures that tend to benefit all road users (e.g. enforcement of lower
speed limits, drunk driving crackdowns, red light cameras, etc).
Though there is no particular reason to suppose that adult pedestrian
and bicycling fatalities should necessarily track each other over time
(and in the US they definitely do not, with adult pedestrian fatalities
declining by 22% since 1975 while adult cycling fatalities have actually
increased by 79% over the same period, despite helmet use rates rising
from <1% to at least 30-40%), there is considerable evidence that such
fatalities for juveniles are very strongly correlated, at least in the
absence of an enforced MHL.
In Great Britain over the period 1979-2004, the correlation
coefficient between cycling and pedestrian fatalities under the age of 16
is an impressive 0.96, and in the US over the period 1975-2004 the same
correlation coefficient is an even more impressive 0.99--strongly
suggesting that these trends have moved in virtual lockstep with each
other over at least the past 25-30 years.
Despite the fact that helmet usage among juvenile cyclists in both
countries has increased over the past two decades from virtually nill to
15-20% in GB and at least 30-40% in the US (where mandatory but seldom
enforced helmet laws apply only to younger cyclists at the state or local
level in about half the states), juvenile cycling fatalities have actually
declined by a slightly smaller proportion than juvenile pedestrian
fatalities in both countries (77.3% vs 80.5% since 1979 in GB, and 78.2%
vs 80.9% since 1975 in the US)[1],[2].
Even if we were to ignore the head injury data from Australia and
elsewhere and agree that bike helmets might still be substantially
effective in preventing potentially fatal head injuries, the failure to
observe any such beneficial effect in either short-term or long-term time
series fatality data at the state or national level must imply that
helmeted cyclists are still managing to get themselves killed at roughly
similar rates to their formerly unhelmeted counterparts--presumably
through some process similar to risk compensation.
An interesting example of why mandating certain safety measures may
not necessarily produce the expected benefits is the related issue of
motorcycle helmets. Unlike the case with bicycle helmets where there is
still considerable debate as to the degree of their actual effectiveness
in preventing potentially serious or fatal head injury, there is
compelling empirical evidence that motorcycle helmets do indeed provide
substantial protection against such head injuries.
Based on the several hundred crashes each decade that involve a
motorcycle operator with a passenger in which one was wearing a helmet and
the other was not, and where one suffered fatal injury while the other
survived, NHTSA estimates motorcycle helmets are about 37% effective in
preventing fatalities overall[3]. There is also compelling evidence
derived from NHTSA's SDS data[4] that the _apparent_ effectiveness of
motorcycle helmets varies enormously depending on whether a particular
state has a MHL that applies to motorcyclists of all ages and/or
experience levels--and if so, how strictly it is enforced.
In those states lacking a universal MHL, and where motorcycle helmet
usage rates are generally 40% or less, the apparent effectiveness of
motorcycle helmets (AE = (FH/FNH)*(NFNH/NFH), where FH=fatal helmet user,
FNH=fatal no helmet, NFNH=nonfatal no helmet, NFH=nonfatal helmet user)
tends to be in line with the empirical evidence, ranging from ~30-50%
(i.e. AE = ~0.7-0.5). Likewise in those states with a strictly enforced
universal MHL where motorcycle helmet usage rates are at least 90%.
In those states with a less strictly enforced MHL however, where
helmet use rates are lower than 85%, apparent effectiveness declines
dramatically and actually turns negative below ~80% usage. This means
helmet-wearing motorcyclists are apparently 20-100%+ more likely to be
involved in a fatal crash than bare-headed motorcyclists in these states
(primarily FL, GA, MD, MI, and PA). Likewise, there were huge swings in
apparent effectiveness in the two states that switched between a less
strict MHL and a non-MHL regime (FL-2000) or vice-versa (MD-1992).
Since the empirical effectiveness of motorcycle helmets presumably
does not vary significantly from one state to the next based merely on
whether a universal MHL is strictly enforced or not, the most plausible
explanation would seem to involve some serious risk compensation effects
among motorcyclists who would not normally choose to wear a helmet if they
were under no legal compulsion to do so. This is supported by evidence
derived from the FARS query system[5] that helmeted motorcyclists are
substantially more likely to be involved in high-speed crashes than bare-
headed motorcyclists, unlike the case with belted vs unbelted drivers of
passenger motor vehicles.
The reason we do not seem to see these risk compensation effects in
either non-MHL or in strictly enforced MHL states is most likely that
selective recruitment (or non-recruitment) effects tend to dominate in
these states. In other words, where motorcyclists are free to choose
whether to wear a helmet or not, generally only the more safety-conscious
choose to do so; and where a universal MHL is strictly enforced, generally
only the most reckless will refuse to wear a helmet. It is only where a
less strictly enforced MHL allows more than the most reckless anti-helmet
minority of motorcyclists to consider riding without a helmet that we
begin to see the full range of risk compensation effects in action.
References:
[1] Fatality data for the US is from FARS http://www-
fars.nhtsa.dot.gov/
[2] Fatality data for Great Britain is from Stats19
http://www.dft.gov.uk/stellent/groups/dft_transstats/documents/divisionh...
[3] http://www-nrd.nhtsa.dot.gov/pdf/nrd-
30/NCSA/RNotes/2005/809861.pdf
[4] http://www-nrd.nhtsa.dot.gov/pdf/nrd-
30/NCSA/Rpts/2002/809_301/15safetyequip.pdf
[5] http://www-fars.nhtsa.dot.gov/queryReport.cfm?stateid=0&year=2004
Competing interests:
None declared
Competing interests: No competing interests
The question asked by the editors of the BMJ is whether enforced
bicycle helmet laws improve public health. Both Robinson (1) and Hagel, et
al (2) provide reasons for their respective views. While the reasons
provide worthwhile and important theories for further research, and should
not be dismissed lightly by those who disagree with them, it is the
original question which is the important consideration in deciding public
policy. In deciding whether or not to legislate, it does not matter why
legislation is or is not effective. What matters is whether or not it
accomplishes what it is intended to do. If it cannot do that, then efforts
are wasted in pursuing it.
The response to Robinson's article seems to imply that the onus is on
those who have not been convinced of the effectiveness of legislation to
prove the reasons they believe legislation would be ineffective. In
reality, the onus is on those who promote legislation to demonstrate that
it will deliver the promised results. In the past, this would have been
difficult because there was no legislation in place for which to observe
results. In the present, there are plenty of examples of legislated
environments to observe. Robinson has demonstrated that legislation has
not had the promised effect in these environments. The response to
Robinson's article seems to argue more about the effectiveness of helmets
than of legislation. It does not demonstrate that legislation is
effective. In fact, it seems to avoid the question. It is troubling that
the authors have clouded the issue by introducing arguments to the effect
that a potential benefit of legislation is to discourage cycling and that
exercise, cycling in particular, is not related to physical fitness.
(1) Robinson, D.L., "No clear evidence from countries that have
enforced the wearing of helmets", BMJ 2006;332;722-
doi:10.1136/bmj.332.7543.722
(2) Hagel, B., A. Macpherson, F.P. Rivara, B. Pless, "Arguments
against helmet legislation are flawed", BMJ 2006;332;725-726-
doi:10.1136/bmj.332.7543.725
Competing interests:
None declared
Competing interests: No competing interests
As someone from outside the medical profession but who is well
accustomed to analysing scientific and legal evidence, I have always been
surprised at the rigid bounds put on evidence deemed acceptable to judge
the effectiveness of cycle helmets. A key principle of sound scientific
evaluation is to seek out and explain apparently confounding results, no
matter from where they come. One of the Cresswell principles of expert
evidence is always to consider material facts which could detract from
one's concluded opinion. In helmet research the extensive real-world
evidence that shows little benefit from helmet wearing has been almost
completely ignored and analysis restricted to a small number of non-
randomised case-control studies despite the widely recognised
susceptibility of these to influences beyond the control of the
researchers.
With experience in investigating cycling crashes I know very well
that the risk of a crash, let alone one involving head injury, is far from
equal across the cycling population but dependent to a considerable degree
upon social class, cycling skill, temperament, where the person is riding
and other variables that are probably impossible to adjust for. The
decision whether or not to wear a helmet and, if so, how one is worn, only
adds to the complexities of comparing those who do and do not suffer head
injuries. It is therefore no surprise to me that Robinson's analysis based
on the real world differs so markedly from the simplistic academic view of
Hagel et al.
I am also surprised that it is deemed acceptable for meta analyses to
be dominated by an analyst's own work, such as is the case with the
Cochrane Review on helmet effectiveness which has influence much beyond
its scientific merit. Given the implications for possible accusations of
bias, prejudice and vested interest, I am amazed that anyone would want to
work under such conditions. In any event, the outcomes of such analyses
are no more robust just because they are based on more than one piece of
flawed research. The public has the right to expect the higher
professional standards of rigour and independence to be found in other
areas of scientific endeavour.
Hagel et al's scepticism about the personal and societal benefits of
cycling is astounding for medical professionals. Is it really concidence
that the countries with the highest levels of obesity are also those with
helmet laws or strong promotion?
Robinson's work and other recent analyses provide the best and most
widely supported evidence yet about the efficacy of cycle helmets. Is it
not time to move on and capitalise on the enormous health benefits that
more cycling would bring? To continue defiantly with recursive references
and extremist claims (such as that helmets prevent 85% of head injuries)
when they have no basis in the real world risks casting shame and ridicule
not just on those making those claims but on the medical profession as a
whole.
Competing interests:
None declared
Competing interests: No competing interests
The Editor, BMJ
Dear Sir,
Following Hagel et al’s opening statement
in their critique of Robinson, it could be said their support for bicycle helmet
legislation is contrary to published evidence of the failure of such legislation
— clearly there is past work with differing conclusions, hence the
debate.
Hagel et al open their critique of Robinson by postulating an
alternative interpretation of Robinson’s Fig 3. We have previously
published [1] an analysis of injury data in New Zealand and a graph based on
that is shown below:
The graph is, unsurprisingly, similar to Robinson’s.
However it differs by showing the change in head injuries for adults and child
bicyclists combined, and also the head injury change for the non-bicyclist
population over the same period. In absolute numbers bicyclist head injuries
only account for a small proportion of the total head injuries. However head
injuries to motor vehicle occupants make up a large proportion of the
total.
The correlation coefficient between the bicycle helmet wearing
rate and bicyclist head injuries is -0.87 (2dp). This agrees with the Hagel
et al’s coefficient of -0.8 for children and -0.9 for
adults. From this they conclude that the benefit of wearing of bicycle helmets
is clear and hence Robinson’s work flawed.
However their conclusion is only obtained by ignoring the
real-world environment that the New Zealand experiment has been
conducted in. We have placed the bicycle data in context by also analysing
the change in head injuries for the population as whole. The coefficient
between the wearing of helmets by bicyclists and the non-bicyclist head
injuries is -0.88 (2dp). If we were to follow Hagel et al’s
hypothesis this would show that to reduce
motorist head injuries you make the bicyclists wear helmets!
Such a conclusion is clearly nonsense. Therefore so also is
Hagel et al’s that the increased bicycle helmet wearing
rate reduced bicyclists head injuries. Indeed the correlation coefficient
between the bicyclist and non-bicyclist head injuries is 0.80, indicating that
the two have tracked each other quite closely, something which a successful
bicyclist-only intervention would change.
Faced with such selective reporting of data it must be Hagel
et al’s argument that “crumbles” and not
that of Robinson.
The New Zealand experiment may not be a case-controlled
study, but it cannot be dismissed as Hagel et al would appear to
wish. A whole population of some 4 million people have been either
compelled to wear helmets while riding a bicycle or told not to bicycle
– despite health indications to the contrary the New Zealand
political message has solidly been that not bicycling is better than doing so
sans plastic hat. The population wide data in New Zealand should show
something like the often claimed, and just as often challenged, 85% reduction
in head injuries to bicyclists. It simply does not.
While some studies, particularly pre-legislation, have shown
some small benefits; other, later studies, have shown total failure. Benefit
cost studies have shown the costs swamping meagre benefits. It might be
nice if the helmet legislation had worked, but wishful thinking doesn’t
prevent injuries or improve health.
It is clear that were the Legislature driven by science and not
political expedience this health, safety & financial disaster would long
have been abandoned.
Yours,
N
Perry, Scientist, New Zealand
[1] Perry, N., The Bicycle Helmet Legislation, Curse or Cure?,
Cycling 2001, Christchurch.
Competing interests:
None declared
Competing interests: No competing interests
The authors' response is very selective in its presentation in order
to press their well known pro-helmet position. It begins by quoting
studies that support their views while curiously omitting any reference to
the many that don't. For example there is no reference to the study by
Rodgers [1] of 8 million cyclists in the USA that found "that the bicycle-
related fatality rate is positively and significantly correlated with
increased helmet use" in line with the findings of Robinson and others for
New Zealand and Australia. Neither is there a reference to the two recent
papers by Hewson [2,3] looking at UK police and hospital statistics which
found no difference in head injury rates in male and female children
despite the helmet wearing rate of the females being double that of the
males. In respect of adults and children Hewson stated "The conclusion
cannot be avoided that there is no evidence from the benchmark dataset in
the UK that helmets have had a marked safety benefit at the population
level for road using pedal cyclists"[3]
The authors reference, in support of their position, six
"independent" and mostly old studies (w1-w6). In one (w2) the authors have
agreed mathematical errors which when corrected revise the conclusion to
every helmet preventing two head injuries - a clear case of confounding if
ever there was one. Two more are the authors' own, one of which [4] is
the source one of the most quoted figures in favour of helmets - an 85%
reduction in head injuries. That study compared helmeted children, mostly
white, well-off, and cycling in parks, with unhelmeted children, mostly
black, cycling in busy urban streets and attributed the difference in head
injury rates to the helmet wearing. It has rightly been criticised for
lacking "scientific rigour" [5]. Yet that flawed figure has become the
rallying cry for much well meaning but misguided helmet legislation and
promotion across the world.
The authors point to Robinson's Fig 2 [6] as evidence of increased
helmet use reducing injury. However even a cursory visual inspection of
Fig 2 shows it would be impossible to identify the year that helmet use
doubled if Robinson had not helpfully marked it for the reader - indeed
most would guess it happened six years earlier in 1985 if not told
otherwise.
The authors dismiss the concept of risk homeostasis in helmet wearers
but conveniently omit reference to their own study that found that injured
children who had worn helmets reported they rode faster and suffered more
damage to their bikes than those without helmets [7]. There is also other
criticism of their "no risk homeostasis" case [8]
Whatever the truth about helmets, cycling is an extremely safe
activity with a lower head injury rate per km than walking. Head injuries
form a lower proportion of all child cyclist hospital admissions in
England (38%) than for child pedestrian admissions (44%). Cyclists
represent just 7% of all hospital head injury admissions exceeded by trips
and falls (42%) and even assaults (11%). Yet while the BMJ had published
many papers on cycling helmets I can only recall one (a tongue in cheek
paper with a serious message by Wardlaw [9]) suggesting helmets for
pedestrians. One has to wonder why cyclists merit singling out by the
medical profession for intervention with helmets. As the Netherlands has
demonstrated, you can achieve the lowest cyclist head injury rate in the
world with helmets being worn by only one in a thousand of your cyclists.
Even so, given the design limitations of cycle helmets (a design
maximum of a stationary fall from a height of 2m onto a flat surface) and
that 97% of cyclist accidents in London involve a motor vehicle it is
clear that this is an intervention with equipment whose design limits fall
well short of the overwhelming majority of cycling injury accident
conditions.
Those who press us down the path of helmet compulsion in the face of
the clearly conflicting evidence risk a repeat of the mistakes made with
hormone replacement therapy [10].
Tony Raven
[1]Rodgers, G.B., Reducing bicycle accidents: a reevaluation of the
impacts of the CPSC bicycle standard and helmet use, Journal of Products
Liability, 1988, 11, 307-317.
[2]Hewson, P.J., Investigating population level trends in head
injuries amongst child cyclists in the UK, Accident Analysis &
Prevention. 2005;37(5):807-815
[3]Hewson P.J., Cycle Helmets and Road Casualties in the UK, Traffic
Injury Prevention, 2005;6(2):127-134
[4] Thompson, R.S., Rivara, F.P. and Thompson, D.C., A case-control
study of the effectiveness of bicycle safety helmets, The New England
Journal of Medicine, 1989 320(21) 1361-1365
[5]Curnow, W.J., 2005. The Cochrane Collaboration and bicycle
helmets. Accid. Anal. Prev. 37 (3) 569-574.
[6]Robinson D.L., No clear evidence from countries that have enforced
the wearing of helmets, BMJ 2006;332:722-725
[7]Mok D, Gore G, Hagel B, Mok E, Magdalinos H, Pless B., Risk
compensation in children’s activities: A pilot study; Paediatr Child
Health. 2004;9(5):327-330.
[8] Adams J, Hillman M, BMJ 2001;322:1063
[9]Wardlaw, M.J., Three lessons for a better cycling future. BMJ:
2000;321:1582-1585
[10]Petitti D. Hormone replacement therapy and coronary heart
disease: four lessons. Int Journal of Epidemiology, 2004;33:461-463
Competing interests:
None declared
Competing interests: No competing interests
In a field as polarised as the efficacy of cycle helmet research it
is hardly surprising that Robinson's work disagrees with a considerable
body of the evidence, and of course exactly the same can be said of pieces
suggesting helmet use is highly beneficial.
While my own reading of the literature suggests to me more holes in
Hagel et als' position than Robinson's, what the pairing of pieces shows
very clearly is that there is still much controversy concerning the
effectiveness of cycle helmets, and this controversy shows no sign of
abating. I cannot see how such an uncertain state of affairs can be any
reasonable basis for a major piece of public health legislation like a
compulsory helmet law for cyclists.
Before any further calls are made for such measures we must take a
much more informed grasp of the true degree of the dangers of cycling. If
one looks at the data (summarised well by Wardlaw, BMJ 2000;321:1582-1585)
we can see that cycling is not particularly more dangerous than being a
pedestrian and those accidents that do happen are not especially more
productive of head injuries. Though a law requiring helmets for
pedestrians might reasonably be assumed to have a similar effect on public
health as one for cyclists we know as a culture happy with the relative
safety of being pedestrians that such a law would be absurd. In countries
that have retained a cycling culture a similar view is evident for
cycling, helmet use amongst cyclists is very low, and so are rates of
cyclist head injuries.
When asking why the UK public (including its legislators, civil
servants, journalists and doctors) has lost its confidence in the safety
of cycling, it appears a highly plausible answer is the extraordinary
amount of time, money and effort spent telling us that cyclists are in
terrible danger so they should wear a helmet. In comparison, pedestrian
safety campaigns do not tell us we are always in terrible danger without a
piece of armour, they tell us we are typically quite safe if we behave
sensibly, which is also true for cyclists. Helmets are not a sensible
answer for pedestrian safety, even though many of them are injured or
killed on the roads every year, and cyclist safety should be approached
the same way: promote skilled interaction with other traffic, and ensure
the other traffic is in turn operating responsibly.
Competing interests:
None declared
Competing interests: No competing interests
1. The original paper shows that the case for cycle helmets is as
strong, or as weak, as for driving helmets or walking helmets. Anyone
seeking to compel cyclists to wear helmets needs to explain why they are
not doing the same to drivers and pedestrians.
2. The critique is starting from the wrong premise: that helmets are
good and it is necessary to prove that helmet compulsion would do harm. It
is actually up to the compulsionists to prove that compulsion would do
good that outweighs the costs it imposes on society.
In trying to minimise the health benefits of regular cycling, the
authors appear to have conceded that helmet compulsion reduces cycling.
Reductions in cycling also cost society in pollution and traffic
congestion, so even if their health argument is valid, it does not follow
that reductions in cycling do not matter.
3. The first paragraph of the critique seems to say that helmets are
very helpful in improving cyclists' safety, but the words used don't. It
says:
"protective association" - not "causal relationship"
"protective effect" - not "reduction in fatal and serious injuries"
"reduction in head injuries" - not "head injury rate reduction"
"reduction in head injuries" - not "greater reduction than in other
comparable groups"
"reduction in head injuries" - not "greater reduction than other
injury types"
The Robinson paper clearly shows that the only changes proven to have
occurred after enforcement of compulsory cycle helmet laws are:
(1) fewer people cycling
(2) a higher proportion of remaining cyclists wearing helmets.
4. Hagel and colleagues appear to believe that questioning Robinson's
suggested reasons why cycle helmets are ineffective (in preventing serious
injury) invalidates the measured casualty data.
The null hypothesis in this situation is "cycle helmets don't protect
heads against serious injuries." Cycle helmet compulsion can only be
justified if this hypothesis can be disproved.
Colin McKenzie
Competing interests:
I do some work as a self-employed cycle trainer for Cycle Training UK, who have a 'helmets optional' policy.
Competing interests: No competing interests
Hagel and colleagues display several simple misconceptions. They
complain that Robinson's work1 is without "concurrent comparison groups".
In fact, Robinson clearly displays the results for several different
control groups including pedestrians, motor vehicle passengers, and
primary school cyclists. Hagel and colleagues supply a calculation from
which they omit any control groups and which comes to the wrong
conclusion. The point is that the proportion of head injuries changes in
parallel in all groups, and shows no effect from the imposition of helmets
on cyclists. Helmet laws demonstrably do not work.
The critics mention the evidence from case-control studies. These
studies are heavily confounded by personal choice. Cyclists who choose to
wear helmets are as a group more cautious than those who do not. They are
involved in fewer accidents with different sorts of injuries. The
resulting biases in case-control studies can probably not be eliminated by
any available statistical adjustment. As an academic exercise, the authors
of these case-control studies may wish to identify the biases if they can.
After Robinson's clear evidence that helmet laws are useless, few others
will wish to spend the time.
Hagel and colleagues seem to be confused about the likely effect of
imposing helmets on people. People who are forced to wear protective gear
are likely to take more risks than they did before. This idea, risk
compensation, is a major issue in many areas, as their own reference makes
clear. It is not the same as the point above, about more cautious groups
tending to adopt "safety" features early.
The health arguments for regular cycling are overwhelming, and helmet
compulsion powerfully discourages cycling2,3. In a large Danish study,
bicycling to work decreased risk of mortality in approximately 40% after
multivariate adjustment, including leisure time physical activity4. Even
sports enthusiasts benefited from regular cycling5. Few interventions have
such a beneficial effect on health. Hagel and colleagues may be right in
saying that occasional gentle leisure cycling has a rather small effect on
health, but it is irrelevant to their argument. Indeed, their point of
view rests on incomprehension and flawed science. Helmet laws do not
reduce the proportion of head injuries among cyclists, and they deter
people from the health benefits of cycling1.
Dr Richard Keatinge
1. Robinson DL. No clear evidence from countries that have enforced
the wearing of helmets. BMJ 2006;332: 722-5
2. Bicycle Helmet Research Foundation.
http://www.cyclehelmets.org/1020.html accessed 26th March 2006
3. http://www.cycle-helmets.com/bicycle_numbers.html accessed 26th
March 2006
4. Andersen LB, Schnohr P, Schroll M, Hein HO. All-Cause Mortality
Associated With Physical Activity During Leisure Time, Work, Sports, and
Cycling to Work. Arch Intern Med. 2000;160:1621-1628
5. Personal communication from Professor Andersen
Competing interests:
Cyclist
Competing interests: No competing interests
Are ecological studies "weak"?
My critics argue that ecological studies are "weak". Yet there is nothing weak about the data showing a 43% reduction in pedestrian fatalities in Victoria about the same time as the helmet law (fig A), nor the large and immediate reduction in fatalities coinciding with the start of random breath testing in New South Wales.
Indeed, the effect of helmet legislation on non-head injuries in Victoria (fig C) is immediate and obvious, although there was little effect on percent head injuries. Could the real "weakness" be that the data don’t show what helmet law advocates want us to believe?
Lessons from epidemiology
There was great concern when randomised control trials (the most reliable form of evidence) showed that hormone replacement therapy increases the risk of heart disease, contradicting an earlier systematic review (including 16 prospective and 11 case-control studies) that concluded it reduced the risk by 50%.[1]
Epidemiologists learned four important lessons. 1) do not turn a blind eye to contradiction, 2) do not be seduced by mechanism, 3) suspend belief and 4) maintain scepticism. Researchers who "believed" could be mistaken.[1]
My critics claim "the effect of increased helmet use is most evident in fig 2".[2] Presumably they mean fig 3, where the correlation with a simple linear trend for both adults and children is 0.97 (r2 0.94). Did "belief" triumph over logic? My critics ignored a much better-fitting model with only a sixth the residual variation, instead claiming helmets were responsible.[2]
My critics also cite studies that "reported a protective association between wearing bicycle helmets and head injuries". Data from one were reanalysed. Helmet wearing increased for girls but decreased for boys. If helmets were responsible, head injury rates should have fallen for girls but increased for boys. Instead, both had an identical declining trend.[3] My critics don’t cite the new evidence, only the original study.
Additional data also contradict the apparent success of legislation in Ontario, Canada. Even at the time, it seemed unlikely that helmets were responsible, because the changes in head injury rates did not coincide with the introduction of legislation.[4] Subsequent surveys showed helmet wearing decreased to pre-law levels by 1999.[5] But head injury rates continued to fall,[6] strongly suggesting the original conclusion was incorrect.
My critics cite papers showing gradual trends in head injury and helmet wearing rates, claiming there is an association. In reality, such studies tell us nothing. It’s a reasonable bet that the correlation between electricity prices in Australia and head injuries in fig 3 was higher than for helmet wearing. But one didn’t cause the other.
To avoid this problem, I focussed on jurisdictions where helmet wearing increased by at least 40 percentage points within a year. I could therefore test a hypothesis: was there any obvious response to the large increase in helmet wearing? The answer was very little in terms of percent head injury, but an obvious effect for non-head injuries in Victoria. Hence my conclusion: that legislation discouraged cycling more than it reduced head injuries.
Health and cycle-use
Obvious benefits of cycling include healthy exercise, reduced greenhouse gas emissions and air pollution (most beneficial if cycling replaces short car trips where the catalytic converter doesn’t have time to warm up), and increased ‘safety in numbers’. My critics dismiss these benefits, claiming that (although a high proportion of respondents in the cited surveys actually said they were deterred by helmet legislation) the decline might "simply reflect increased in-line skating or other recreational activities!" Public health professionals might like to comment.
Onus of proof
The onus of proof should be on those who favour removing the freedom to choose. Claims of huge reductions in head injuries from helmet legislation simply didn’t materialise. So let’s move on to measures such as those that reduced pedestrian fatalities in Victoria by 43% (with similar benefits expected for cyclists). These measures worked. Helmet legislation didn’t. Let’s see how similar campaigns can be used to create a safer, healthier road environment for everyone.
References
1 Petitti D. Commentary: hormone replacement therapy and coronary heart disease: four lessons. Int J Epidemiol 2004;33:461-3.
2 Hagel B, Macpherson A, Rivara FP, Pless B. Arguments against helmet legislation are flawed. BMJ 2006;332:725-726.
3 Hewson PJ. Investigating population level trends in head injuries amongst child cyclists in the UK. Accid Anal Prev 2005;37:807-15.
4 Robinson DL.
Confusing trends with the effect of helmet laws. Pediatrics P3Rs, 2003. (accessed March 2006).
5 Cycling Scotland. Helmet policy statement, associated briefing paper and helmet briefing bibliography 2005. (accessed Dec. 2005).
6 CIHI. Injury Hospitalizations (includes 2000-01 and 2001-02 data). Canadian Institute for Health Information, 2003.
Competing interests:
None declared
Competing interests: No competing interests
Health benefits of utility cycling: evidence overlooked
Hagel et al[1] attempt to rebut Dorothy Robinson's evidence[2] that
helmet laws in Australia have had a serious negative effect on public
health, by deterring cycle use whilst failing to improve cyclists’ safety.
In their response, it appears that they do at least accept the evidence
(summarised at [3]) that laws requiring cyclists to wear helmets do indeed
discourage substantial numbers of people from cycling. However they then
go on to argue this does not necessarily indicate damage to public health,
as it is unclear by how much those people reduce their cycle use, or for
how long.
Hagel et al cite a "rule of thumb" (derived from an item of
unpublished personal correspondence) that one needs to cycle for 45
minutes a day, 6 days a week to gain health benefits, and then suggest
that few leisure cyclists are doing this much, implying (presumably) that
the health benefits that we might stand to loose through helmet compulsion
are pretty meagre. Surely though it is not right to refer solely to
“leisure cyclists”, when their own “rule of thumb” suggests that the best
way to gain health benefits is in fact by riding regularly as part of
one’s day-to-day routine – for journeys to work, to school or college, to
the shops? A two-way cycle commute of 20-30 minutes each way is not
unusual, and would easily cover the 1 hour or 40km per week which they
cite as being necessary to reduce CHD risks in 45 to 64 year olds. In any
case, there is other published evidence that cycling for as little as 3km,
3 days a week can still provide benefits[4].
Given that only 35% of men and 24% of women are meeting the national
recommendations for physical activity[5], the threat from helmet-
compulsion is not just that it would reduce existing cycle use, but that
it would seriously undermine efforts to encourage more people to cycle
more often. Cycling, together with walking, is one of the few forms of
physical activity (aside from a few minority activities: jogging, inline
skating etc) which can readily be undertaken in the course of ordinary
daily travel. Unlike gym attendance or most "sporting" activities,
cycling does not require any special outlay of either time or money, and
can readily be fitted into ones normal routines. Once adopted, it is a
habit which is easy to maintain. Moreover, cycle use provides the
greatest benefits in other policy areas besides health – e.g. reduced
congestion, air pollution, road danger and greenhouse emissions – when
undertaken not as a leisure activity but as way of reversing the
inexorable growth of unnecessarily car use, especially for short journeys.
A comprehensive literature review of the health benefits of
cycling[6] provides references to many similar findings, several of them
suggesting that there is a “dose-response” relationship between cycle use
and health gains, but that low levels of cycle use are still beneficial.
A key reference however is a population-wide study from Copenhagen which
found that, compared with people who cycled regularly to work, those who
did not do so had a 39% higher mortality rate, regardless of any other
cycling or other physical activity undertaken by those in each group[7].
This indicates that cycling provides very substantial reductions in all-
cause mortality, which in turn suggests that its health benefits must far
outweigh the risks involved.
That was indeed what the BMA found in its report on Cycling and
Health[8]. Other evidence indicates that those who cycle into middle
adulthood can have a level of fitness equivalent to being 10 years
younger[9] and a life-expectancy 2 years above the average[10]. The
author of the BMA report later estimated that, thanks to these extra life-
years, the health benefits of cycling outweigh the risks by a factor of
around 20:1[11].
When we learn that, two years after helmet laws were introduced,
there was a 43% reduction in cycle use among under 16 year olds in New
South Wales[12] and a 46% reduction among teenagers in Melbourne[13], it
surely does not take a full epidemiological study to realise that those
teenagers growing up after the law are vastly less likely to cycle in
later life than their predecessors (in any case, I dread to think how one
could isolate the effects of a helmet law in a long time-series study of
the type Hagel et al are suggesting, or what it would cost to conduct it).
What is clear is that we need to find ways to support increased cycle use,
both to encourage children to retain the cycling habit through their
teenage years and into adulthood, and indeed for adults to rediscover
cycling, so that exercise becomes part of their daily norm. Cycle
training is a good way to encourage this[14], whereas helmet-compulsion
clearly has exactly the opposite effect.
Hagel et al also fail to note that heart disease due to physical
inactivity kills 42,000 people a year[15], that obesity shortens the lives
of 30,000 people by an average of 9 years per person[16], and that the
costs of physical inactivity and obesity are £8.2bn and £2.5bn
respectively[17]. The numbers killed while cycling are tiny by comparison
– typically around 130 p.a. in recent years[18]. Even then, it is unknown
how many of these could be prevented by helmet-wearing, however helmets
are only designed for impact speeds equivalent to falling from a
stationary riding position[19], whereas around 90% of cyclists’ fatal and
serious injuries involve colissions with motor vehicles[20], making it
very unlikely that helmets could prevent many of the fatal or really
serious injuries which occur, even on the most optimistic assumptions
about their effectiveness. Indeed, we do not know how many of the
cyclists killed are wearing helmets at the time, although media reports
often mention that a helmet was in fact being worn.
In short, helmet laws have the evident potential to shorten hundreds,
if not thousands more lives than they could ever hope to save. Their
effect is not only to reduce existing cycle use but also to deter new
people from becoming regular cycle users. Cycle use for day-to-day travel
is rare – just 1.5% of trips in Britain are cycled[21]. Yet Britain still
has relatively high levels of recreational cycle use – around 9 million
adults (aged over 16) in Britain cycle at least once a year and about half
of them cycle at least once a week[22]. Given concerns not only about
obesity and heart disease, but also congestion, air quality and climate
change, we need to encourage more people to adopt the habit of cycling as
a regular activity, and the large numbers of occasional recreational
cyclists (including children and teenagers) are the obvious “target
group”. We need to achieve rates of cycle use similar to those of
continental neighbours such as Sweden, Germany, Denmark or the Netherlands
(countries which incidentally have much lower helmet-wearing rates, and
better cycle safety – see graph at www.cyclehelmets.org.uk). The vast
majority of Britain’s current adult and child cyclists do not wear
helmets, particularly when riding on minor roads[23]. The last thing we
should be doing is criminalising them any further into car-dependent,
sedentary lifestyles.
Roger Geffen
Campaigns & Policy Manager
CTC, the national cyclists’ organisation
REFERENCES
[1] Hagel B et al. “Arguments against helmet legislation are
flawed.” BMJ vol 332 pp725-726, 2006.
[2] Robinson DL. “No clear evidence from countries that have
enforced the wearing of helmets.” BMJ vol 332 pp 722-5, 2006.
[3] Bicycle Helmet Research Foundation. “How helmet promotion
affects cycle use.” www.cyclehelmets.org/mf.html?1021.
[4] Blair S et al. “Changes in Physical Fitness and All-Cause
Mortality: A prospective study of healthy and unhealthy men.” Journal of
the American Medical Association 1995; vol. 273 pp 1093-98, 1995.
[5] Joint Health Surveys Unit. “Health survey for England 2004 –
updating of trend tables to include 2004 data.” The Stationary Office,
2004.
[6] Cavill N & Davis A. “Cycling and health: a briefing paper
for the Regional Cycling Development Team.” ERCDT, 2003.
[7] Andersen L et al. “All-cause mortality associated with physical
activity during leisure time, work, sports and cycling to work.” Archives
of Internal Medicine vol. 160 pp 1621-8, 2000.
[8] British Medical Association. “Cycling: towards health and
safety.” Oxford University Press, 1992.
[9] Tuxworth W et al. “Health, fitness, physical activity and
morbidity of middle aged male factory workers.” British Journal of
Industrial Medicine vol 43. pp 733-753, 1986.
[10] Paffenbarger R et al. “Physical activity, all-cause mortality
and longevity of college alumni.” New England Journal of Medicine, vol.
314(10) pp 605-613, 1986.
[11] Hillman M. “Cycling and the promotion of health.” PTRC 20th
Summer Annual Meeting, Proceedings of Seminar B, pp 25-36, 1992.
[12] Smith N & Milthorpe F. “An observational survey of law
compliance and helmet wearing by bicyclists in New South Wales.” New South
Wales Roads and Traffic Authority, Sydney, 1993.
[13] Finch F et al. “Bicycle use and helmet wearing rates in
Melbourne, 1987 to 1992: the influence of the helmet wearing law.”
Accident Research Centre report no. 45, pp 35, 36, 43. Monash
University,1993.
[14] Cycle Training UK. “Survey on the effectiveness of cycle
training.” CTUK 2004.
[15] Britton A & McPherson K. “Monitoring the progress of the
2010 target for coronary heart disease mortality: estimated consequences
on CHD incidence and mortality from changing prevalence of risk factors. A
report for the Chief Medical Officer.” National Heart Forum, 2001.
[16] National Audit Office. “Tackling obesity in England.”
Stationary Office, 2001.
[17] Department of Health. “At least five a week: evidence on the
impact of physical activity and its relationship to health.” DoH, London
2004.
[18] National Audit Office. “Tackling obesity in England.”
Stationary Office, 2001.
[19] Glanville H and Harrison N. “Cycle helmets.” British Medical
Association, 1999.
[20] Department for Transport. “Road Casualties Great Britain 2004.”
Table 23. DfT, 2005.
[21] Department for Transport. “Transport statistics Great Britain
2005” Table 1.4. DfT, 2005.
[22] Fox K & Rickards L. “Sport and leisure: Results from the
sport and leisure module of the 2002 General Household Survey.” Office for
National Statistics, 2004.
[23] Inwood C et al. “Cycle helmet wearing in 2004”. Report TRL 644
for the Department for Transport, Transport Research Laboratory, 2005.
Competing interests:
None declared
Competing interests: No competing interests